Professional Documents
Culture Documents
ABSTRACT
The molecular structure of benzene has been determined by combining the average
distances obtained by the present electron diffraction study and the moments of inertia
reported by Cabana et al. The following thermal average bond distances have been deter-
mined: rg(C-C) = 1.399 + 0.001 A and r,(C-H) = 1.101 -c 0.005 A. The uncertainties
represent the estimated limits of error. The C-H distance of this molecule is similar to
vinyl C-H distances.
INTRODUCTION
It is often found that the structural data of the most basic molecules are
not very up-to-date because such molecules were studied in the early stages
of the development of experimental methods for structure determination.
Benzene is one such molecule.
The molecular structure of benzene in the gas phase was determined by
means of electron diffraction [l-3] and Raman spectroscopy [4, 51; the
values determined by the previous electron diffraction studies, ra(C-C) =
1.397 i- 0.004 A and r,(C-H) = 1.08 f 0.02 a, have often been quoted as
the interatomic distances of this molecule. Since these values were obtained
more than a decade ago, we thought it worthwhile to reinvestigate the
structure of benzene using modem experimental and analytical facilities. In
the midst of our present study, an electron diffraction reinvestigation of
benzene has also been performed by Seip et al. [6,7] and Beagley and
Hewitt [S] , independently. In view of the importance of benzene,
however, we have completed the structure determination using the latest
spectroscopy results in addition to those we obtained by electron diffraction.
The combined use of diffraction and spectroscopic data has been success-
ful in the determination of accurate structures of gas molecules [9-111. In
the present study, bond lengths were determined independently by electron
diffraction and utilisation of spectroscopic results, then combined use of
them was made in order to obtain the most reliable values of the structural
parameters.
In the initial stage of the present study, we used the rotational constants
244
ELECTRON DIFFRACTION
Experimental
*Walker et al. have recently redetermined the rotational constant, B,, of carbon disulfide
by means of laser Raman spectroscopy [ 143. By using their constant, the r=(C-S)
distance was calculated to be 1.5569 a. We have so far used the value, r,(C-S) = 1.5570 A
[15], which was determined from the B, value observed by Stoicheff [16]. Fortunately,
the difference between the two r,(C-S) values is so small that it has no effect upon the
results of our previous structure determinations.
**The experimental total intensity data and error matrices have been deposited with the
British Library Lending Division at Boston Spa, Yorkshire, U.K. as Supplementary
Publication No. 26015 (9 pages).
245
10
10 20 30 40
r-C%)
Fig. 2. Observed radial distribution curve for benzene and the residuals. Vertical lines
mark interatomic distances.
TABLE 1
rc-c ‘C-H
*The sources of systematic errors taken into consideration were as follows: uncertainty
involved in the determination of the scale factor, uncertainty in the parameter, K,
uncertainty in the correction curve for the sector shape, delocalization of the sample gas,
arbitrariness of the background curves and arbitrariness of the weight function.
247
Fig_ 3. Molecular intensities for benzene. Theoretical curves calculated by the use of the
best-fit parameters and the residuals, sM(.s)ObS - SMUT
*Our calculated frequencies agreed well with those reported by Caiifano and Crawford
except for the E,,v,, mode of vibration of benzene-d,, for which we obtained a frequency
higher by 30 cm-‘. The reason for this discrepancy is not known. However, this mode of
vibration makes only an insignificant contribution to the mean amplitude_ Therefore, no
further refinement of the force constants was made.
248
TABLE 2
AI,b
aDerived from the B, values observed by Cabana et al. 1121 by using the conversion
factor, 16.8575 amu A2 cm-‘.
bCorrection for the vibrational effect.
%ero-point moments of inertia, It,(‘) = fdeff) + AI,.
dValues in the parentheses are uncertainties in 10m3amu AZ units.
TABLE 3
GH, 43 15
GD, 42 13
Conversion of rg into ri
The rz distance can be derived from the rg distance by using the anhar-
manic stretching, <Ar>, the perpendicular mean amplitudes at zero
temperature, KO, and the centrifugal distortion, 6 rcent[ 101. The centrifugal
terms were estimated by balancing the centrifugal force with Hooke’s
restoring forces of the C-C and C-H bonds. The numerical values for this
*The definitions of various average distances, rz, rg and rav, are given in refs. 30 and 31.
249
TABLE 4
<Pr> KS ’ rcent
r -r"
K Q
CC 1 24 2 27(3)=
C-H 0 164 0 164(6)
TABLE 5
b
‘g
C-C 1.395, 0.001, 1.396, 0.000, 1.396, 1.399, 0.000,
C-H 1.091 0.012 1.083 0.006 1.085 1.101 0.005
conversion are listed in Table 4. The ?-i structures of benzene thus derived
are listed in the first column of Table 5 along with their limits of error
which were estimated from the errors of the rEdistances and those involved
in the conversion.
When D,, symmetry is assumed for the present molecule, the b-moment
of inertia can be expressed in terms of the two parameters, r, (C-C) and
r,(C-H), by the equation,
Ibcz)= 3{m,(rZCC)z + mH(rzcC + rzcEz)2} (1)
The rz structure of this molecule can, therefore, be determined from the
spectroscopic data alone by using the estimated values of the isotopic
differences in the rz distances. From the observed I,,(=) values of C6Hs and
C,D, in Table 2 and the isotopic differences in Table 3, the r, values were
determined as shown in Table 5.
Figure 4 shows the C-C and C-H distances obtained and their limits of
error. In Fig. 4 a spectroscopically allowed range for each species is confined
by the parallel lines which represent the limits of error. The width of the
parallel line allowed from Ib(Z)(CsD6) was determined fkom the errors of the
1b~z)(C6D,) and those involved in the isotopic differences. The parallelogram
enclosed by the solid lines, therefore, indicates the region of the parameter
allowed by the spectroscopic data. The uncertainties involved in the a values
250
l.lOC
1.09-
Q-=x
z
L
c 1oe-
1 1
1.394 1396 1390
t-K-C)(~)
Fig. 4. Relation between r(C-C) and r(C-H). A parallelogram shows a region of values
allowed by the two moments of inertia and an ellipse shows the limits of error of the
diffraction data-The points ED and SP represent the values obtained by electron diffrac-
tion and spectroscopy, respectively. The straight line shows the direction of possible
variation of rz(C-C) and rg(C-H) when a systematic error is allowed for in the scale
factor of the diffraction data.
I
1.396 1397
Tz (C-C 1 (E 1
Fig. 5. The rz values obtained using various sets of a values. (1) e&c = 2-l and cC__H =
2.0 K’; (2) cc_ = 2.1 and q-H = 2.5; (3) acs = 2.1 and“C__H = 1.5; (4) acS = 2.6
andac_H = 2.0; (5) ocx = 1.6 and eC__H = 2.0.
were assumed to be + 0.5 tS_’ for both the C-C and C-H bonds. The in-
fluence of the a values upon the rz structure is illustrated in Fig. 5. The
small circles show the r,yalues obtained by using several sets of a values
which were taken within their assumed uncertainties. As may be seen, the a
value for the C-H bond has much greater influence upon the results than
does that for the C-C bond. The errors due to the uncertainty in the 0: value
251
unount to about one-half as much as the quoted limits of error for the
results.
The most reliable values of the distances were determined by the combined
use of the r: distances and the moments of inertia. The two parameters
r,(C-C) and r,,(C-H) were adjusted so as to minimize the weighted sum
of the squares of residuals,
R = F W, {rE(obs) - r,” I:, + z, IV,. {1,<=’(obs) - Ib e) (talc) ) ;:n (2)
where the summations were taken over the two distances and the two
moments of inertia. The relative weighting factors for the observed values
were given as being inversely proportional to the squares of the limits of
error. The weighting factor for Ib w (&De), however, was estimated by adding
the uncertainty in the isotopic difference to the limit of error. The results are
shown in Table 5. The limits of error of these parameters were estimated from
the limits of error of the observed values by applying the law of propagation
of errors. The moments of inertia calculated using the structural parameters
thus obtained agree well with the observed values as shown in Table 6.
DISCUSSION
*Diehl and Niederberger have concluded from the agreement between their NhIR spectro-
scopic result and our preliminary electron diffraction result that there is no detectable
change in the interatomic distances on changing from the gas phase to the liquid crystal
solvent. However, it has been found finally that the C-H distance in our early stage of
the present study was somewhat too long.
252
TABLE 6
ACKNOWLEDGEMENTS
REFERENCES